PartiqleDTR

Repository of the Partiqle Decay Tree Reconstruction (DTR) project. The PartiqleDTR project attempts to tackle the problem of particle decay tree reconstruction using hybrid quantum-classical machine learning approaches.

🚀 Setup and Run

1. Clone the repository
2. Install dependencies
   • Option A: Poetry installed on your system

   poetry install
   

   • Option B: Create a venv, and install dependencies from pyproject.toml manually
3. Create a dataset and run training using the existing parameters:

   kedro run
   

🔧 Configuration

The command prefix poetry run can be omitted if you're using an already activated venv

• Pipeline Configuration:
  • Open visualization: poetry run kedro viz
  1. Data Generation:
     • Builds decay tree
     • Generates decay events
  2. Data Processing
     • Preprocessing events
     • Generating datasets
  3. Data Science
     • Model creation
     • Training
• Parameters can be adjusted in the following locations:
  • conf/base/parameters/data_generation.yml
  • conf/base/parameters/data_processing.yml
  • conf/base/parameters/data_science.yml
• Runs are being recorded using MLFlow
  • Open dashboard: poetry run mlflow ui

Notes

torch_scatter needs gcc-c++ and python3-devel packages to build successfully.

with poetry you need to release the tensorflow-probability dependency from phasespace such that the line in .venv/lib/python3.10/site-packages/phasespace-1.8.0.dist-info/METADATA becomes

Requires-Dist: tensorflow-probability (>=0.15)

To achieve this, you may want to run

poetry add phasespace

prior to the installation of the other packages using

poe

changed module dependencies in phasespace:

before:

Requires-Dist: tensorflow (<2.8,>=2.6)
Requires-Dist: tensorflow-probability (<0.14,>=0.11)
Requires-Dist: keras (<2.7)

after:

Requires-Dist: tensorflow (>=2.6)
Requires-Dist: tensorflow-probability (>=0.11)
Requires-Dist: keras (<2.10)

then 
pip install -U tensorflow should update keras, numpy etc
pip install zfit (uninstall first if already installed manually)


### MLflow

upgrade kedro
run kedro mlflow init